ES2207824T3 - PROCEDURE AND SYSTEM FOR AUTOMATIC ACQUISITION OF FORM DATA. - Google Patents

Abstract

The invention relates to a method and an arrangement for the automatic data acquisition of paper-based information, whereby an unknown form (10) that is scanned into a computer is transformed into line elements (14, 15) for quicker identification. The identification of known forms (10) is also simplified by means of the line elements (14, 15).

Description

Procedure and automatic acquisition system of forms data.

Technical field

The present invention relates to a method and to a device for automatic data acquisition, by means of of means for this, of forms whose design and information are not know in advance, by entering these media along with the pattern storage of them.

Prior art

A problem of companies, organizations and others is to make good use of the information found in various types of paper forms, documents, etc.

With the new modern technology, these items they can be scanned with a scanner and incorporated into a database through software programs available in the market. Without However, classification, identification and other routines of checking should still be done largely manually through the computer screen.

For example, to store an invoice of an e identical company as a document specifically designed with a logo and other visual elements, must be revised to adapt its format to another that can be accepted by the software and then stored in a database. This and other procedures are they must repeat each time an invoice is scanned with the software With a new design.

To identify an invoice from a company that You have already registered, you often look for the entire invoice, which is a slow operation Some software programs may have routines of search that restrict the degree of such search. Nevertheless, it is difficult to safeguard against veiled or handwritten lines of text, etc.

Therefore, there is a general need for Quickly handle invoices and other forms to be able to identify them and / or be able to quickly incorporate and store new patterns in your billing system.

US-A-4 933 979 describes the traditional acquisition of form data and requires predefined templates / patterns without the ability to self-learning (adaptive). The stored boxes Form defaults are used to identify the form.

UA-A-5 Patent 140 650 mentions the acquisition of form data with what it is known as "form" technology to cover the original document and retain only the parts "completed." This data acquisition is often combined with the acquisition of data according to US-A-4 933 979. This Patent does not It has no adaptive function for data acquisition of unknown forms.

Other patent, US-A-5 293 429, refers to the classification of documents with the help of lines in documents and does not refer directly to the acquisition of the data or any adaptive function for it. US-A-5 293 429 does not guarantee the identification of lines with areas of objects (areas with text) and an "RCG Value" (recognition, a number that identifies only one document).

None of these patents generates a map of form for a form unknown to the system according to the patent and store the map in real time in a database of forms for recognition of the next opportunity to ID. Therefore, for inventions according to these patents, the unknown form must then be stored by other media.

Summary of the Invention

One of the objectives of the present invention is solve the problems indicated above as well as others during which is called automatic data acquisition (interpretation) in connection with paper based information management.

The present invention relates to a system (method and device) for automatic data acquisition of forms where the system has no previous knowledge the appearance of the form or where the form is to be Find the information. In this way, the templates of forms do not have to be defined in advance, but rather recorded instead when submitted to the system, that is, in real time.

To achieve these objectives, this invention specifies a method and a device for acquisition automatic data, by means of means, of forms whose design and information is not known in advance, by the entry into said media together with the storage of patterns of them. The method is adaptive, so it includes learning and registering forms as patterns without text filled in, and also including the following steps to achieve the adaptive registration:

generation of a form map based on the unknown form design in advance to identify the information contained in the form;

Search and compare the form map with the stored maps, registered on media to store maps of form;

store form maps generated in the storage media when they don't match a stored map according to predetermined limits for concordance;

indication of agreement according to limits of agreement when concordance is found; Y

continuous data acquisition for Identification of the information content of the form.

According to an embodiment of the present invention, the form map can consist of a list of zones of objects with the objects contained in the form so the Object includes colors and / or totally or partially text.

In an alternative embodiment, the map of form constitutes a line map that includes objects in Form color lines form.

The horizontal lines on the line map are used to produce a horizontal key by dividing the form in a predetermined number of horizontal segments along the axis and in a coordinate mapping system, so each segment is equivalent to a position in the horizontal key.

The vertical lines in the line map are used to produce a vertical key by dividing the form in a predetermined number of vertical segments along the x axis in a coordinate map system, so each segment is equivalent to a position in the vertical key.

At least one line item that is included in a segment it is marked in the equivalent key position, and the segments that lack line items remain unmarked in The equivalent key position.

A horizontal key and / or a vertical key they constitute a line key in the line map, so during said search, the generated line key is compared to line keys stored in the media to verify the agreement.

Line keys are classified in the media of storage according to the number of brands.

The horizontal position of the object in the list of Object zones are used to generate a horizontal key dividing the form into a predetermined number of segments horizontal along the axis and in a cartographic system of coordinates, so each segment is equivalent to one position in the horizontal key.

The vertical position of the object in the list of Object zones are used to produce a vertical key dividing the form into a predetermined number of segments verticals along the x axis in a cartographic system of coordinates, so each segment is equivalent to one position in the vertical key.

At least one object that is included in a segment is marked in the equivalent key position, and the segments that lack objects remain unmarked in the equivalent key position.

A horizontal key and / or a vertical key they constitute an object key in the list of object zones, so, during that search, the generated object key is compared to the object keys stored in the media for Verify the agreement.

Object keys are classified preferably in the storage media according to the number of brands.

The search results in a predefined number of Probable candidates requested for the form currently wanted.

If necessary, an operator can support manually all or parts of the adaptive record or of the identification of the new form or registered forms respectively if several alternative candidates are found as odds according to a merit factor.

Finally, the identity of the form is confirmed by the acquisition of data of an RCG value.

In addition, the present invention specifies a device to perform said method.

The device performs the acquisition automated data, through means to do so, forms whose design and information is not known in advance, by the entry into said device in conjunction with the storage of their patterns. Learn adaptively and record the Form design, and includes a computer with the media Following to perform adaptive registration:

means for generating a map of form based on the design of the unknown form by in advance to identify information contained in the form;

means to search and compare the map of form with stored maps, registered in media for store form maps;

means to store form maps generated on storage media when they do not match a map stored according to match limits default;

means to indicate agreement according to limits of agreement when concordance is found; Y

means for identification and acquisition continued data content information from the form.

In addition, the device may include or constitute as specified according to said method of the present invention, which is best illustrated in the accompanying dependent claims of the device.

Brief description of the drawings

Additional reference to the included figures and the associated text will give a clearer understanding of this invention.

Figure 1 schematically shows how to Make a line pattern from a scanned invoice.

Figure 2 schematically shows a route of flow to scan, identify, interpret and validate a form according to the present invention.

Detailed description of preferred embodiments

In the following description of this invention, the forms are presented as invoices. Nevertheless, The invention is not limited to invoice forms, but It also covers general documents containing text, figures, etc. as forms Invoices are used here as an example of a form to exemplify the invention.

Figure 1 schematically illustrates a part of invoice 10 that is scanned on a computer and displayed on the screen. As is evident from invoice 10, it is unclear or Blurred after scanning or entry.

Invoice 10 consists in part of a logo 12 and elements of vertical lines 14 and horizontal 15.

Note that even the logo contains 16 long black or variable color tone linear elements that have been partly registered on a line map 18 according to the present invention, and that give an idea of the appearance of the logo original 12, which simplifies identification when the invoice it is an object that will be identified as registered in a database of form map data. The color lines also include Gray scales in black.

The form map that in this case constitutes line map 18 has been filtered from other objects 19, such as all or parts of text objects or color objects, more said line elements that include color, which cannot be reproduce here, but you can include many color fields in a form 10.

An invoice 10 prepared in accordance with this invention, which hereafter referred to as EH (Eyes & Hands), is due identify early. For the identification to succeed, EH must have learned on a previous occasion the aspect of the current bill 10, which in simple terms means that the Information about the invoice is available on the basis of Forms data in EH.

The identification must be necessarily fast and be able to be done in a database that contains a very number large bill identities 18. It is not uncommon for databases contain more than 10,000 identities 18.

The method and device that uses EH does not It requires always passing an invoice through a scanner exactly of the same way, that is, the invoice information may vary something on the x and y axes within a certain measure or a threshold value  predetermined. Figure 1 shows a cartographic system coordinate schematic

In the present invention, (EH) includes that, in one embodiment, EH look for in the invoice all the elements of vertical lines 14 and horizontal 15 of a length default Lines 14, 15 do not have to be independent nor be isolated, but, for example, can be part of a 12 larger logo text, such as ReadSoft AB in the figure 1. Logo 12 is represented as linear element 16 in the line map 18.

Horizontal lines 15 and lines verticals 14 constitute the basis for generating a horizontal key (key h) and a vertical key (key v) respectively as next:

*

The invoice is divides a large number of horizontal segments along the y axis (not shown). Each segment is equivalent to a position in the key h. If a certain segment includes one or more line elements 15, a mark or indicator is placed in the key position equivalent. If not, an empty space is used instead, a inverted brand or anything that distinguishes itself from a brand.

*

It generates a key v for vertical line elements 14 similarly along the x axis.

*

The h keys and v receive designations and together constitute a key to line. After this, a search is performed, which means that the current line key is compared to the line keys of 10 known invoices that exist in the EH database. Is comparison considers that individual lines or elements of line 14, 15 may vary some position, and that the total pattern of the lines may be somewhat offset according to values convenient defaults in the x and y directions, horizontal and vertically, respectively.

*

The keys of line in the database are classified according to the number of brands (indicators), which are used to make the search effective.

*

The search results in a predetermined number of probable candidates for the identity of the current invoice 10. All candidates are associated with a merit factor or a probability of being the current invoice 10.

*

The identity of the invoice is finally confirmed by making an interpretation of what is known as the RCG value (RCG-recognition). The RCG value is a value in a given position that is unique to a certain invoice / vendor or other form Examples of such values are the turn numbers of bank, turn numbers of the post office, numbers of invoice, total amounts, etc.

Such segments may form, for example, checkered patterns that are screened to varying degrees according to the Relative need for quick search.

Line keys can even be implemented in objects formed totally or partially of text and colors. They are assign line keys from a list of object zones that includes the x and y keys for the object. The list of zones of objects can consist, for example, of positions for certain Selected objects The principles for line maps above are even suitable for objects other than Line elements to achieve the identification of forms.

If the line keys are not found in the database, this indicates that the invoice is not known, which gives place for new line keys to be stored at the base of data that, in this way, is updated in real time.

If necessary, the operator can, through its computer, manually support all or part of the adaptive registry and / or the identification of a new form or a form registered, respectively, if there are several alternative candidates likely according to the merit factor.

In addition, the present invention includes a device to perform the method as indicated.

The device performs automatic acquisition of data, through means to do so, of the forms whose Design and information content is not known in advance, by the entry into said media together with the storage of patterns of them. Register in an adaptive way and learn Form design, and includes a computer with the media Following to achieve adaptive registration:

means to generate a form based map in the design of the unknown form in advance to identify information contained in the form;

means to search and compare the map of form with the recognized maps stored in media for store form maps;

means for storing maps of form generated on storage media when not match a map stored according to match limits default;

means to indicate agreement according to limits of agreement when concordance is found; Y

means for identification and acquisition continued data content information form.

Said means are preferably controlled by hardware and software, for example:

A scanner for data acquisition.

An electronic storage medium (disk hard, CD-ROM, etc.) for the means of storing the information

Signs, icons, signal generators, etc., for Indicate purposes.

Filters and comparators so that the media can search and compare, as well as filters and records to ID.

Collectively, the means used herein invention are known to those skilled in the art, but the way they coordinate to achieve the object of the invention It is, however, innovative.

In an embodiment of the present invention referring to figure 2, a schematic route of flow to show scanning, identification, interpretation, and the validation of a form according to the present invention.

Figure 2 is divided by dashed lines in partial zones to clarify the various steps of a method according to the invention, where the steps constitute the scanning of the form 200, the identification of form 210, the interpretation of form 220, and form validation 230.

The form is scanned 200 in EH, and follows the identification 210. Identification consists in generating a map of lines 212, or alternatively from a list of object areas, by what a line key is generated. After this, the form 10 214 is compared with known keys in the map database of form, so you get a conformation of the Identification using the RCG value. The next step includes decide if the identification was successful 216 according to the conditions "Yes or no". If the decision is "No", a conditional investigation to see if there are more candidates in the line code form 218. If the answer is "Yes", it loop through forms 214, 216 and 218 until finally a successful identification is made, or until it is not presents no other candidate of line keys 218.

In the case of a successful identification, it Start interpretation 220 of the form with the help of the map current form 222, after which the validation 230 or evaluation 232 of the fields of the form 10. As an option, the operator can contribute to the selection if they find several alternative fields 234.

If identification 210 is unsuccessful, and does not presents no other line code 218, interpretation is made 220 in which self-study is carried out with a definition of form 224. The definition of the form consists of a template or set of rules that describes the elements of common fields of a specific collection of forms, by example, Swedish invoices. After that, the value is interpreted RCG 226 and a decision 228 is made on whether the current RCG value is You can find it in the forms database. Yes the answer it is "Yes", a reinterpretation 229 begins, followed by a Continuous interpretation 222 leading to validation 232.

On the other hand, if the answer is "No", validation 230, 236 begins, after which the form it is saved in the form map database with the key of line 238. Before steps 236, 238, the operator can, if there is several field alternatives, contribute to the process of self-learning

The embodiments of the present invention before described are possible embodiments, but it is not intended to limit the invention to them, since other embodiments will be apparent to those skilled in the art by the exposed in the attached claims.

Claims (32)

1. Method for the automatic acquisition of data (200), by means of this, of forms (10) whose design and content of information (19) is not known in advance, by the entry in said media, together with the storage of the patterns of the same, where the method is adaptive (224) and includes self-learning and registration of the design of forms without defining form templates in advance, characterized in that it includes the following steps to perform adaptive registration (238) : generation of a form map (18) based in the design (10) of the unknown form in advance to identify (210) information contained in the form; search and compare (210, 220) the map of form (18) with the registered maps stored in some means for storing form maps; storage (238) of form maps generated (18) on storage media when they do not match with a map stored according to match limits default; indication of agreement according to limits of agreement when concordance is found; Y continuous data acquisition (232) for Identify the information content of the form. 2. Method according to claim 1, characterized in that the form map (18) consists of a list of areas of objects with the objects (19) contained in the form. 3. Method according to claim 2, characterized in that the object (19) includes colors and / or totally or partially text. Method according to claim 1, characterized in that the form map (18) constitutes a line map that includes elements of lines (14, 15) of the form (10). 5. Method according to claim 4, characterized in that the horizontal lines (15) in the line map are used to generate a horizontal key by dividing the form into a predetermined number of horizontal segments along the axis and in a coordinate mapping system , so each segment is equivalent to a position in the horizontal key. Method according to claims 4 and 5, characterized in that the vertical lines (14) in the line map are used to generate a vertical key by dividing the form (10) into a predetermined number of vertical segments along the x-axis in a cartographic coordinate system, so each segment is equivalent to a position in the vertical key. Method according to claims 5 and 6, characterized in that at least one line element (14, 15) that is included in a segment is marked in the equivalent key position, and because segments that lack line elements remain Unchecked at the equivalent key position. Method according to claims 4-7, characterized in that the horizontal key and / or a vertical key constitute a linear key in the line map (18), whereby, during said search, the generated line key is compared ( 214) with the line keys stored in the storage media to verify the match. 9. Method according to claim 8, characterized in that the line keys are classified in the storage media according to the number of marks. Method according to claims 1 and 2, characterized in that the horizontal position of the object in the list of object areas is used to generate a horizontal key by dividing the form into a predetermined number of horizontal segments along the axis and in a system coordinate map, so each segment is equivalent to a position in the horizontal key. Method according to claims 1 and 2, characterized in that the vertical position of the object in the list of object areas is used to generate a vertical key by dividing the form into a predetermined number of vertical segments along the x axis in a system coordinate map, so each segment is equivalent to a position in the vertical key. Method according to claims 10 and 11, characterized in that at least one object that is included in a segment is marked in the equivalent key position, and because the segments that lack objects remain unmarked in the equivalent key position. 13. Method according to claims 10-12, characterized in that a horizontal key and / or a vertical key constitute an object key in the list of object areas, whereby, during said search, the generated object key is compared with the object keys stored in the storage media to verify the match. 14. Method according to claim 13, characterized in that the object keys are classified in the storage media according to the number of marks. 15. Method according to claims 1-14, characterized in that the search (210, 218) results in a predefined number of probable candidates requested for the currently sought form. 16. Method according to claims 1-15, characterized in that the identity of the form is confirmed by the acquisition of data of a recognition value (RCG) (214) that identifies only one form. 17. Device for the automatic acquisition of data, through means for this, of forms (10) whose design and information content (19) is not known in advance, for the entry into said media together with the storage of patterns from them, it learns adaptively and registers the design of the forms (10) without defining form templates in advance, and includes a computer, characterized by the following means to perform adaptive registration (238): means to generate a form map (18) based on the design (10) of the unknown form in advance to identify the information contained in the form; means to search and compare the map of form with stored maps recognized in the media for store form maps; means for storing maps of generated form (18) on storage media when not match a map stored according to match limits default; means to indicate agreement according to limits of agreement when concordance is found; Y means for identification and acquisition continued data content information form. 18. Device according to claim 17, characterized in that the form map (18) consists of a list of areas of objects with objects (19) contained in the form (10). 19. Device according to claim 18, characterized in that the object (19) includes colors and / or totally or partially text. 20. Device according to claim 17, characterized in that the form map (18) constitutes a line map including line elements (14, 15) of the form. 21. Device according to claim 20, characterized in that the horizontal lines (15) in the line map are used to generate a horizontal key by dividing the form into a predetermined number of horizontal segments along the axis and in a cartographic system of coordinates, so each segment is equivalent to a position in the horizontal key. 22. Device according to claims 20 and 21, characterized in that the vertical lines (14) in the line map are used to generate a vertical key by dividing the form into a predetermined number of vertical segments along the x-axis in a cartographic system of coordinates, so each segment is equivalent to a position in the vertical key. 23. Device according to claims 21 and 22, characterized in that at least one line element (14, 15) that is included in a segment is marked in the equivalent key position, and because segments that lack line elements remain not marked in the equivalent key position. 24. Device according to claims 21-23, characterized in that the horizontal key and / or a vertical key constitute a line key in the line map, whereby, during said search, the generated line key is compared with the keys of line stored in the storage media to verify the concordance. 25. Device according to claim 24, characterized in that the line keys are classified in the storage media according to the number of marks. 26. Device according to claims 17 and 18, characterized in that the horizontal position of the object (19) in the list of object areas is used to generate a horizontal key by dividing the form into a predetermined number of horizontal segments along the y axis. in a coordinate mapping system, so each segment is equivalent to a position in the horizontal key. 27. Device according to claims 17 and 18, characterized in that the vertical position of the object (19) in the list of object areas is used to generate a vertical key by dividing the form into a predetermined number of vertical segments along the x-axis in a cartographic coordinate system, so each segment is equivalent to a position in the vertical key. 28. Device according to claims 26 and 27, characterized in that at least one object that is included in a segment is marked in the equivalent key position, and because the segments that lack objects remain unmarked in the equivalent key position. 29. Device according to claims 26-28, characterized in that a horizontal key and / or a vertical key constitute an object key in the list of object areas, whereby, during said search, the generated object key is compared with the Object keys stored in the storage media to verify the match. 30. Device according to claim 29, characterized in that the object keys are classified in the storage media according to the number of marks. 31. Device according to claims 17-30, characterized in that the search results in a predefined number of probable candidates requested for the currently searched form. 32. Device according to claims 17-31, characterized in that the identity of the form is confirmed by the acquisition of data of a recognition value (RCG) (214) that uniquely identifies a form.